ENVI Experiment Tutorial (5) Experiment 5, Remote Sensing Image Enhancement

Experiment 5. Remote sensing image enhancement

1. Purpose of the experiment

5. Master the principles and method steps of spatial domain, radiation domain and spectral domain enhancement;
6. Master the principles and methods of band combination.
7. Master the basic image enhancement operations such as ENVI histogram stretching, spatial convolution operation, spectral transformation, band operation and color synthesis.

2. Basic requirements of the experiment

3. Carefully read and master the content of this experiment.
4. Save and record the experimental results, and analyze and summarize.
5. Clear steps and corresponding results (figures or tables, etc.) are required in the experiment report.

3. Experiment time and place

3. Place:
4. time:

4. Experimental conditions

1. Hardware: PC computer (Windows operating system)
2. Software: ENVI 5.3
3. Reference material: Chapter 5 of "ENVI Remote Sensing Image Processing Method"
4. Use data: D:\Program Files\Exelis\ENVI51\data\ qb_boulder_msi
   /...\Chapter 5 Image Enhancement

5. Experimental content

1. Radiation Domain Enhancement Processing
2. Spectral Domain Enhancement Processing
3. Spatial Domain Enhancement Processing

6. Matters needing attention

1. When performing spatial filtering operations, adjust the window size to analyze its impact on the filtering results;
2. Download a remote sensing image of the Landsat-8/OLI Poyang Lake area in the geospatial data cloud. Try different band combinations and analyze which combination can better identify water bodies, vegetation, and buildings?
3. No matter what kind of augmentation operation, try to think about the need and purpose behind it.

Seven, the main steps of the experiment

1. Radiation (contrast) enhancement
(1) Gray level threshold segmentation
(1) Open the "qb_boulder_msi" data.
(2) Right-click on the "qb_boulder_msi" layer in the Layer Manager (Layer Manager) and select the New Raster Color Slices menu. Select a band of the image in the file selection dialog box, click the OK button to open the Edit Raster Color Slices panel

(3) In the Edit Raster Color Slices panel, there are two ways to perform grayscale segmentation:

1.     自动分割
   

2.     手动输入分割区间
   

(2) Contrast stretching
Select the Stretching Data tool in the Toolbox, select the data to be stretched, and the stretching method selection interface will pop up

Use the stretch drop-down menu in the toolbar, select the layer to be stretched in the layer manager on the left, and then pull down to select the stretch method

ENVIclassic Operation steps: Select the Enhance menu in the main image window—>Interactive Stretching to enter the interaction stretched interface.

Linear (linear stretching)
piecewise Linear (segmented linear stretching)
Gaussian (Gaussian stretching)
Equalization (histogram equalization stretching)

2. Spectral Domain Enhancement
(1) Multi-spectral Band Four Arithmetic Operations
• NDVI Calculation Tool
Find /Spectral/Vegetation/NDVI tool in Toolbox, open and set the corresponding sensor band


• Band Math tool
• 1. Must conform to IDL language to write band operation expressions
• 2. All input bands must have the same space size
• 3. All variables in the expression must be named with Bn (or bn)
• 4. The result band must have the same space size as the input bands
• 5. Call the IDL written When customizing a function,
the band calculation tool can call the Function written in IDL. When the function is a source code file (.pro), ENVI+IDL must be started to call it; if the function is compiled into a sav file, you can put the sav file in the following path and restart ENVI can be called.
• ENVI 4.x: C:\Program Files\ITT\IDL\IDL80\products\envi48\save_add
• ENVI Classic: C:\Program Files\Exelis\ENVI51\classic\save_add
• ENVI 5.x: C:\Program Files\Exelis\ENVI51\extensions

3. Example of band calculation usage
   (1) Start ENVI, select the menu File > Open, and open the data "can_tmr.img";
   (2) Start the Band Math tool, the path is Toolbox/Band Ratio/Band Math;
   (3) In the Band Math panel, Enter the calculation expression in the Enter an expression text box: b1+b2+b3, click the Add to List button to add the expression to the Previous Band Math Expression list; (4) In the
   Band Math panel, select the added "b1 +b2+b3", click the OK button to open the Variables to Bands Pairings dialog box (as shown in the figure), and assign image files or image bands to each variable in the calculation expression (5) In the
   
   
   Variables to Bands Pairings dialog box, Variables used in Select the variable b1 in the expression list box, and click "TM Band 1 (0.4850)" in the Available Bands List. Then use the same method to specify "TM Band 2" and "TM Band 3" for b2 and b3;
   (6) Click the Choose button, select the file name and path to save the result, and click the OK button to execute the operation.
   (7) At this point, the input and output files can be loaded into the view, and then click the toolbar icon to get the cell value of the current mouse position
   

(2) Spectral feature transformation

3. Image space filtering
(1) Convolution filtering

(2) Mathematical Morphological Filtering
(3) Texture Analysis

8. Specific requirements of the experiment
1. Select different stretching methods to stretch the same image, and compare the stretching effects of different methods.

2. Please use the NDVI tool to calculate the normalized normalized water index NDWI of the fast bird image (qb_bonlder_msi) that comes with ENVI5, and output and compare the two images of NDVI and NDWI.

（NDVI）

(NDWI)
3. Please use the band calculation tool to realize, calculate the normalized difference vegetation index, and output the result.

4. Referring to the operation steps in 5.3 of the textbook P115-121, perform color transformation, ICA transformation, MNF transformation, PCA transformation, TC transformation and color stretch transformation on the same multispectral image (optional), and finally use color transformation on the transformed result. Composite output and comparative display.

（pca）

（ICA）

(color space transformation)

decorrelation stretching

Photographic Stretch

(saturation stretch)

(mnf)

(tc)
5. Referring to the operation steps in 5.1 of the textbook P106-111, carry out spatial filtering, morphological filtering and texture analysis on the same DEM data (cutting a small area from the world elevation data GMTED2000 with large terrain fluctuations). A spatial domain enhancement method selects two kinds of filtering windows (3 3, 7 7) respectively for processing, and finally outputs and compares the enhanced results.

8. Experimental results and discussion
Gray-scale segmentation can separate ground objects very well, but it also has limitations.
Through different spatial domains and spectral domains, you can get the information you want.